Clinical Chemistry 58:2

391401 (2012)

General Clinical Chemistry

Specificity Characteristics of 7 Commercial Creatinine

Measurement Procedures by Enzymatic and Jaffe
Method Principles
Neil Greenberg,1* William L. Roberts,2 Lorin M. Bachmann,3 Elizabeth C. Wright,4 R. Neil Dalton,5
Jack J. Zakowski,6 and W. Greg Miller3

BACKGROUND: Standardized calibration does not change a

creatinine measurement procedures susceptibility to
potentially interfering substances.
METHODS: We obtained individual residual serum or
plasma samples (n 365) from patients with 19 different disease categories associated with potentially interfering substances and from healthy controls. Additional sera at 0.9 mg/dL (80 mol/L) and 3.8 mg/dL
(336 mol/L) creatinine were supplemented with acetoacetate, acetone, ascorbate, and pyruvate. We measured samples by 4 enzymatic and 3 Jaffe commercially
available procedures and by a liquid chromatography/
isotope dilution/mass spectrometry measurement procedure against which biases were determined.

implementation of a method principle influenced its susceptibility to potential interfering substances.

2011 American Association for Clinical Chemistry

CONCLUSIONS: There were differences in both magnitude

and direction of bias among measurement procedures,
whether enzymatic or Jaffe. The influence of interfering
substances was less frequent with the enzymatic procedures, but no procedure was unaffected. The details of

The 2006 report from the National Kidney Disease

Education Program (NKDEP)7 Laboratory Working
Group (LWG) (1 ) highlighted the need for improved standardization of routine measurements of
serum and plasma creatinine (2, 3 ). Although calibration was emphasized in the report, interferencerelated bias was not overlooked. In recommendations for in vitro diagnostics (IVD) manufacturers,
the report stated, IVD manufacturers must address
analytical nonspecificity bias in current routine serum creatinine methods.
Standardization of creatinine measurement procedures with calibrations traceable to isotope dilution mass
spectrometry (IDMS) reference measurement procedures has largely been accomplished for major manufacturers in North America, as evidenced in external quality
assessment schemes using commutable samples. For example, in the February 2010 College of American Pathologists (CAP) Creatinine Accuracy Calibration Verification/Linearity (LN24) Survey, 11 method groups had
mean bias of 0% (range 5.8% to 7.7%) vs NIST reference measurement procedure values for creatinine concentrations between 0.77 mg/dL (68 mol/L) and 4.09
mg/dL (362 mol/L), and the CV for all participants (n
372) was between 6.6% and 2.9% (4 ).
The NKDEP LWG and the International Federation
of Clinical Chemistry and Laboratory Medicine Working
Group on Glomerular Filtration Rate Assessment designed the present study to obtain contemporary data for

RESULTS: The number of instances when 3 or more results

in a disease category had biases greater than the limits of
acceptability was 28 of 57 (49%) for Jaffe and 14 of 76
(18%) for enzymatic procedures. For the aggregate group
of 59 diabetes samples with increased -hydroxybutyrate,
glucose, or glycosylated hemoglobin (Hb A1c), the enzymatic procedures had 10 biased results of 236 (4.2%)
compared with 89 of 177 (50.3%) for the Jaffe procedures,
and these interferences were highly procedure dependent.
For supplemented sera, interferences were observed in 11
of 24 (46%) of groups for Jaffe and 8 of 32 (25%) of
groups for enzymatic procedures and were different at
low or high creatinine concentrations.

Ortho Clinical Diagnostics, Rochester, NY; 2 University of Utah, Salt Lake City,
UT; 3 Virginia Commonwealth University, Richmond, VA; 4 National Institute of
Diabetes and Digestive and Kidney Diseases, NIH, Department of Health and
Human Services, Bethesda, MD; 5 WellChild Laboratory, Kings College London,
Evelina Childrens Hospital, London, UK; 6 Beckman Coulter, Inc., Brea, CA.
* Address correspondence to this author at: 40 Chipping Ridge, Fairport, NY
14450. E-mail ngreenbe@frontier.com.
Received July 5, 2011; accepted October 11, 2011.
Previously published online at DOI: 10.1373/clinchem.2011.172288

Nonstandard abbreviations: NKDEP, National Kidney Disease Education Program; LWG, Laboratory Working Group; IVD, in vitro diagnostics; IDMS, isotope
dilution mass spectrometry; CAP, College of American Pathologists; VCU,
Virginia Commonwealth University; ARUP, Associated Regional University Pathologists; DCMP, designated comparison measurement procedure; LC-IDMS,
liquid chromatography/IDMS; SRM, Standard Reference Material; Hb A1c, glycosylated hemoglobin; JCTLM, Joint Committee for Traceability in Laboratory
Medicine; eGFR, estimated glomerular filtration rate; IFU, instructions for use.

391

use in establishing specificity performance recommendations for serum creatinine measurement procedures.
Materials and Methods
SAMPLES EXAMINED

Individual patient samples. The study was approved by

the respective institutional review boards at both clinical sample collection locations: Virginia Commonwealth University (VCU) and University of Utah Associated Regional University Pathologists (ARUP)
laboratories.
Sample handling and storage conditions were consistent with published information on stability of creatinine in human serum or plasma (5 ). Individual patient serum or heparinized plasma samples were
collected in accordance with standard laboratory practices. Residual samples submitted for laboratory testing
were stored up to 8 h at room temperature then up to
14 days at 2 4 C before division into aliquots and frozen storage at 70 C (VCU) or up to 24 h at room
temperature then up to 30 days at 20 to 30 C before thawing, aliquoting, and refreezing at 70 C
(ARUP). Five 0.25-mL aliquots, in 1.2-mL cryovials
(VWR International) were prepared from each residual sample and stored at 70 C until being shipped
on solid CO2 to participating manufacturers and the
designated comparison measurement procedure
(DCMP) laboratory. Samples were stored at 70 C
at each participating laboratory until being measured
between January and March 2009.
We selected samples to obtain up to 20 individual
serum or plasma samples in each of 19 disease categories established to have either known concentrations
of, or a high probability to contain, substances suspected as potential interferents in creatinine measurement procedures (1 ). Samples were not pooled, except
those with increased pyruvate were obtained from
pooled whole blood that was incubated at room temperature 24 48 h before centrifugation. We identified
the selected samples in each category (Table 1) on the
basis of previously measured laboratory values, patient
location consistent with 1 or more of the selected disease categories, or medical record review. In addition,
for a control group, we collected samples from 20 nondiseased individuals (laboratory staff volunteers) who
were not using any prescription or over-the-counter
pharmaceutical or dietary supplement products.
Samples spiked with potential interfering substances. Serum with a creatinine concentration of 0.90 mg/dL (80
mol/L) was obtained from a healthy male volunteer, age 48 years. Half was supplemented with a solution of 150 mg/dL (13.26 mmol/L) creatinine hy392 Clinical Chemistry 58:2 (2012)

drochloride (Sigma Aldrich) dissolved in water to a

final creatinine concentration of 3.80 mg/dL (336
mol/L). Candidate interfering substances were
added as described in the Supplemental Data, which
accompanies the online version of this article at
http://www.clinchem.org/content/vol58/issue2.
INSTRUMENTS AND MEASUREMENT PROCEDURES

Designated comparison measurement procedure. We

measured serum or plasma creatinine by liquid chromatography/IDMS (LC-IDMS) based on the method
described by Preiss et al. (6 ). Procedural details are
provided in the online Supplemental Data.
Commercial clinical laboratory measurement procedures. Participating manufacturers and respective measurement
procedures were as follows: Beckman Coulter Synchron
Unicel enzymatic (E1) and Jaffe (J1), Ortho Clinical Diagnostics Vitros 5,1 FS Chemistry System enzymatic (E2),
Roche Diagnostics Integra enzymatic (E3) and Jaffe (J2),
and Siemens Dimension RxL enzymatic (E4) and Jaffe
(J3). Characteristics of the commercial creatinine procedures, and details of the protocol for preanalytical sample
handling and creatinine measurement at the participating manufacturers laboratories, are provided in online Supplemental Table S1.
Characterization of additional analytes. Procedures for
measurement of concentrations of additional analytes
used for selection and characterization of patient samples
were as described in Table 1, footnote a.
STATISTICAL PROCEDURES

For each commercial procedure and the DCMP, we

used the mean of quadruplicate creatinine measurements on each sample for all statistical analyses and
comparisons.
For samples supplemented with interfering substances, we calculated percentage recovery as the creatinine concentration at each concentration of the interferent divided by the creatinine concentration of the
unspiked sample, multiplied by 100. Percentage bias
was the percentage recovery minus 100. We used standard linear or polynomial regression to estimate the
relationship between percentage recovery of creatinine
and the concentration of an interferent.
For individual patient samples, we calculated bias
as the difference from the DCMP creatinine concentration. On the basis of the distribution of biases observed
for the samples from healthy controls (Fig. 1, A and B),
the criteria for presence of a nonspecificity bias was
defined as the larger of 0.10 mg/dL (8.8 mol/L) or
10%. For the rationale for these criteria, see online Supplemental Data: Additional Statistical Considerations.

Specificity of Creatinine Measurement Procedures

Table 1. Patient sample selection criteria for potential interfering substances.a

Category

Median
creatinine,
mg/dL (range)b

Selection criteria

Patient samples in the category

20 samples from 14 patients, 4 of whom had 3,
3, 2, and 2 different samples each
19 samples from 19 patients

0.82 (0.305.27)

20 samples from 5 patients, 2 of whom had 10

and 7 different samples each

2.23 (0.473.01)

20 samples from 20 patients

1.35 (0.564.48)

20 samples from 19 patients, 1 of whom had 2

different samples
20 samples from 20 patients

0.76 (0.479.04)

Dobutaminee

Serum/plasma concentrations 1.44.0 mg/dL

(median 2.0 mg/dL)
Serum/plasma concentrations 33103 mg/dL
(median 64 mg/dL)
Siemens Advia index 4; serum/plasma
concentrations 9.3 mg/dL for 1 sample and
15.637.1 mg/dL (median 32.5) for the remaining
samples
Outpatient cardiac care clinics; with hypertension
and taking 3 or more HTN/CVD medicationsc
Patients receiving cefepime, ceftriaxone, cefazolin,
cefoxitin, or cefpodoximed
Blood collected immediately before hemodialysis
procedure
Patients receiving dobutamined

1.96 (0.662.96)

Dopaminee

Patients receiving dopamined

eGFR, low

eGFR 1530 mL min1 (1.73 m2)1

Glucose, high

Serum/plasma concentrations 388816 mg/dL

(median 455)
Plasma Hb A1c concentrations 8.1%13.2% (median
8.9%); glucose concentrations 77461 mg/dL
(median 261)
Siemens Advia index 4; approximate serum/plasma
hemoglobin concentrations 3501000 mg/dL
(median 450)
Patient status posttransplant, taking 1 or more
immunosuppressant drugs (tacrolimus, sirolimus,
cyclosporine, mycophenolic acid)
Serum/plasma concentrations 0.310.8 mg/L
(median 4.3)
Siemens Advia index 23, approximate
serum/plasma triglycerides 500-1000 mg/dL based
on intralipid equivalents
Serum/plasma concentrations 6.917.9 mg/dL
(median 10.4)
Serum/plasma concentrations 3.16.2 mg/dL
(median 4.2)
Urine albumin 226 and 547 mg/L (n 2), albumincreatinine ratio 2983 mg/g (n 1), urine protein
between 3.1 and 21.8 g/L (n 15)
Pooled heparinized blood at room temperature for
2448 h, plasma concentrations 0.190.34 mmol/L
(median 0.29)

18 samples from 11 patients, 6 of whom had 3,

2, 2, 2, 2, and 2 different samples each
11 samples from 6 patients, 3 of whom had 3,
3, and 2 different samples each
19 samples from 18 patients, 1 of whom had 2
different samples
20 samples from 18 patients, 2 of whom had 2
different samples
20 samples from 20 patients

Albumin, low

-Hydroxybutyrate
Bilirubin, high

Cardiovascular
disease
Cephalosporins
Dialysis

Hb A1c, high

Hemolyzed

Kidney transplant

Lidocaine
Lipemic

Protein, high
Protein, low
Protein, urine,
high
Pyruvate

0.89 (0.2210.11)

5.38 (1.5011.33)

2.19 (0.552.61)
3.02 (2.154.24)
1.10 (0.342.83)
0.83 (0.642.03)

20 samples from 18 patients, 1 of whom had 2

different samples, and 1 of whom had a second
sample in the high urine protein group
20 samples from 20 patients

0.88 (0.468.35)

20 samples from 20 patients

0.79 (0.496.10)

20 samples from 15 patients, 2 of whom had 4

and 3 different samples each

1.39 (0.214.99)

20 samples from 20 patients

0.99 (0.511.88)

20 samples from 20 patients

1.02 (0.344.72)

18 samples from 15 patients, 2 of whom had 2

different samples, and 1 of whom had a
second sample in the hemolyzed group
20 different pooled samples

2.94 (0.977.95)

1.31 (1.002.94)

1.16 (0.772.28)

Albumin, glucose, and total protein in serum/plasma samples were measured by use of a Vitros 5,1 FS Chemistry System (Ortho Clinical Diagnostics).
-Hydroxybutyrate was measured by use of Liquicolor reagent (Stanbio Laboratory) on a Cobas c501 analyzer (Roche Diagnostics). Bilirubin, hemolysis, and lipemia
indexes were estimated by use of a Siemens Advia 1650 analyzer (Siemens Medical Solutions Diagnostics). Bilirubin concentrations were measured by use of the
Siemens Advia procedure. Approximate hemoglobin concentrations were based on index values from Ortho Vitros and approximate triglyceride values on index
values from Siemens Advia. eGFR was calculated with the IDMS traceable MDRD Study equation with creatinine measured by use of Roche enzymatic Creatinine
Plus reagents and Calibrator for Automated Systems adapted to a Siemens Advia 1650 analyzer. Hb A1c was measured by use of phenylboronic acid affinity HPLC
procedure (Trinity Biotech), lidocaine by use of an Abbott TDx analyzer (Abbott Laboratories), urine albumin and protein by use of a Siemens Advia 1650 analyzer,
and pyruvate by use of a spectrophotometric procedure with lactate dehydrogenase and NADH (19 ).
b
Measured by LC-ID-MS/MS.
c
Hypertension (HTN)/cardiovascular disease (CVD) medications included diuretics, vasodilators, calcium channel blockers, -blockers, angiotensin-converting
enzyme inhibitors, angiotensin II receptor antagonists, 3-hydroxy-3-methylglutaryl-coenzyme Areductase inhibitors (statins), or other cholesterol-reducing agents
and platelet aggregation inhibitors.
d
Samples were obtained during the drug-dosing interval when patients were expected to be at steady state.
e
Two patients received both dopamine and dobutamine.

Clinical Chemistry 58:2 (2012) 393

Fig. 1. Creatinine measurement bias vs DCMP for individual serum or plasma samples from the apparently healthy
control group (A and B), the hemolyzed group (C and D), and the delayed separation group (E and F).
Note that 1 sample with creatinine 8.35 mg/dL is not shown in C but had biases of 0.10, 0.02, 0.28, and 0.56 for
procedures E3, E4, E2, and E1, respectively, nor in panel D but had biases of 0.20, 0.20, and 0.14 mg/dL for procedures J1, J2,
and J3. Symbols: , procedure E1 or J1; , procedure E2 or J2; E, procedure E3 or J3; , procedure E4. The lines indicate the
criteria used for a meaningful nonspecificity bias. Multiply by 88.4 to convert creatinine to mol/L.

Results
CALIBRATION TRACEABILITY

We assessed trueness for each creatinine measurement

procedure by recovery of the certified concentration with
394 Clinical Chemistry 58:2 (2012)

or without expanded uncertainty for NIST Standard Reference Material (SRM) 967 Creatinine in Frozen Human
Serum (7 ) (online Supplemental Fig. S3). The DCMP
method results (mean and expanded uncertainty) were
0.751 (0.029) mg/dL [66.4 (2.6) mol/L] and 3.850

Specificity of Creatinine Measurement Procedures

Table 2. Maximum percentage bias observed for serum supplemented with potential interfering substances.
Creatinine measurement procedure, maximum bias, %
Substance

E1

E2

E3

E4

J1

J2

J3

Acetoacetate 174 mg/dL

Creatinine 0.9 mg/dL

1.9

2.1

0.5

0.2

43.9

27.2

Creatinine 3.8 mg/dL

30.9

0.1

11.3

5.5

4.7

0.6

0.7

Creatinine 0.9 mg/dL

1.9

0.1

0.8

0.8

28.3

20.5

30.3

Creatinine 3.8 mg/dL

7.5

0.8

8.3

2.2

0.1

0.6

122

Acetone 100 mg/dL

Ascorbate 20 mg/dL
Creatinine 0.9 mg/dL

12.4

Creatinine 3.8 mg/dL

2.5

11.6

9.6

7.7

86.1

177.4

5.7

13.3

42.3

5.4

6.8

0.2

0.5

54.7

40.2

50.2

0.4

7.3

0.8

1.7

0.4

Pyruvate 1.2 mmol/L

Creatinine 0.9 mg/dL
Creatinine 3.8 mg/dL

1
14.1

(0.081) mg/dL [340.4 (7.2) mol/L] for the 2 concentrations, and were within the SRM uncertainty intervals. In
addition, Community Bureau of Reference Certified Reference Materials 573, 574, and 575, Human Serum, were
included in all runs for the DCMP. See online Supplemental Tables S2 and S3 for additional data on the calibration traceability of the DCMP.
For the commercial procedures, bias was 5.9%
for SRM 967-1 and 3.8% for SRM 967-2. Four of 7
means were within the uncertainty interval of SRM
967-1 (see online Supplemental Fig. S3A). The 2 SD
ranges for 2 means overlapped the uncertainty interval,
and the 2 SD range for 1 mean (procedure E4) was
just outside the uncertainty interval. For SRM 967-2
(see online Supplemental Fig. S3B), 3 of 7 means were
within the uncertainty interval, the 2 SD ranges for 3
means overlapped the uncertainty interval, and the 2
SD range for 1 mean (procedure E1) was just outside
the uncertainty interval.
For all commercial procedures, mean biases for
137 of 140 results (98%) in the healthy individuals were
within 0.10 mg/dL (8.8 mol/L) of the DCMP (Fig. 1,
A and B), and the mean biases for each individual procedure ranged from 5.5% to 5.8%, further supporting calibration traceability to IDMS reference measurement procedures (8 ). With either the mean bias with
SRM 967-1 or the mean bias for healthy individuals, in
conjunction with CV estimates based on either the
within-procedure imprecision for SRM 967-1 obtained in
this study or with the within-procedure/amonglaboratories CV from the CAP Comprehensive Chemistry Survey C-B (2011), the total allowable error of all commercial procedures was within the recommendations of
the NKDEP LWG (1 ) (see online Supplemental Fig. S4).

2
11.5

Therefore, we used no corrections for calibration biases in

any of the commercial procedures in the data analysis.
INTERFERING SUBSTANCES SUPPLEMENTED INTO SERUM

Owing to concerns about their stability, acetoacetate,

acetone, ascorbate, and pyruvate were supplemented
into serum with withinreference interval and increased creatinine concentrations. The maximum percent biases observed for each supplemented interferent
at each concentration of creatinine are summarized in
Table 2, and interferographs are shown in online Supplemental Data Figs. S5S8.
INDIVIDUAL SAMPLES FROM CLINICAL CATEGORIES

Difference plots for the biases observed between routine creatinine measurement procedures and the
DCMP are shown for apparently healthy controls and
hemolyzed and delayed sample processing (Fig. 1); increased glucose, increased glycosylated hemoglobin
(Hb A1c), and increased -hydroxybutyrate (Fig. 2);
and low albumin, low total protein, and high total protein (Fig. 3) clinical sample categories. The biases observed for the other clinical sample categories are
shown in online Supplemental Figs. S9 S19.
Table 3 summarizes results for all clinical sample
categories by measurement procedure. For 1 or more
of the Jaffe procedures, 3 biased creatinine values per
clinical sample group were observed for the increased
-hydroxybutyrate, increased glucose, increased
Hb A1c, cardiovascular disease, cephalosporin, dobutamine, lidocaine, increased bilirubin, delayed sample
processing, hemolyzed, lipemic, low albumin, high total protein, and post kidney transplant groups. For 1
or more of the enzymatic procedures, 3 biased creatClinical Chemistry 58:2 (2012) 395

Fig. 2. Creatinine measurement bias vs DCMP for individual serum or plasma samples from the groups with
increased glucose (A and B), increased Hb A1c (C and D), and increased -hydroxybutyrate (E and F).
Note that 1 sample with creatinine 10.11 mg/dL is not shown in panel E but had biases of 0.39, 0.31, 0.48, and 0.96
for procedures E2, E3, E4, and E1, respectively, nor in panel F but had biases of 0.33, 0.02, and 1.27 mg/dL for procedures
J3, J1, and J2. , procedure E1 or J1; , procedure E2 or J2; E, procedure E3 or J3; , procedure E4. The lines indicate the
criteria used for a meaningful nonspecificity bias. Multiply by 88.4 to convert creatinine to mol/L.

396 Clinical Chemistry 58:2 (2012)

Specificity of Creatinine Measurement Procedures

Fig. 3. Creatinine measurement bias vs DCMP for individual serum or plasma samples from the low albumin group
(A and B), the low total protein group (C and D), and the high total protein group (E and F).
, procedure E1 or J1; , procedure E2 or J2; E, procedure E3 or J3; , procedure E4. The lines indicate the criteria used for
a meaningful nonspecificity bias. Multiply by 88.4 to convert creatinine to mol/L.

inine values per clinical sample group were observed

for the increased -hydroxybutyrate, dobutamine, lidocaine, increased bilirubin, hemolyzed, lipemic, and
high total protein groups. Of the 365 clinical samples,
the overall proportion of biased results was 11.7%,
9.0%, 8.8%, and 11.2% for the 4 enzymatic methods

and 19.5%, 15.6%, and 35.9% for the 3 Jaffe methods.

Overall, more biases were observed for the Jaffe procedures, but findings were inconsistent in terms of occurrence rate, direction, and magnitude of bias (Table 3,
Figs. 13, and online Supplemental Figs. S9 S19),
within a method principle (Jaffe or enzymatic).
Clinical Chemistry 58:2 (2012) 397

Table 3. Number of samples in each sample category with a negative () or positive () bias >0.10 mg/dL
(>8.8 mol/L) or >10%, whichever was greater.
Creatinine measurement procedure
E1
Sample category

E2

E3

E4

J1

J2

J3

Apparently healthy

20

Clinical categories

365

31

12

27

21

11

33

70

39

18

128

Diabetes mellitus

-Hydroxybutyrate 33103 mg/dL

19

11

14

Glucose 388816 mg/dL

20

14

19

Hb A1c 8.1%13.2%

20

16

Cardiovascular disease with hypertension

20

Drugs
Cephalosporins

20

10

Dobutamine

18

Dopamine

11

Lidocaine

20

10

11

Bilirubin 938 mg/dL

20

13

10

18

19

Delayed separation 2448 h

20

13

Hemolysis, hemoglobin 350 mg/dL

20c

11

Lipemia

20

Albumin 1.44.0 g/dL

20

Protein 718 g/dL

20

17

Protein 3.16.2 g/dL

20

Predialysis

20

eGFR 1530 mL min1 (1.73 m2)1

19

Postkidney transplant

20

Urine protein 322 g/L (n 15)

18

Endogenous substances

Protein abnormalities

Kidney disease

Discussion
Calibrations of the LC-IDMS DCMP and the commercial measurement procedures were verified to be
traceable to Joint Committee for Traceability in Laboratory Medicine (JCTLM) listed reference measurement procedures and reference materials. Consequently, calibration bias of the commercial
creatinine measurement procedures was not a factor
in evaluating the influence of potential interfering
substances.
The serum creatinine concentration influenced
the magnitude of specific interferences in our supplementation studies. The interferographs, particularly
for the Jaffe procedures, demonstrated substantial in398 Clinical Chemistry 58:2 (2012)

terferences at low creatinine concentrations but no interference at high creatinine concentrations. Nearly all
of the interferences observed for the Jaffe procedures
were positive, with the exception of acetoacetate for
procedure J2. The magnitude of the interference observed for the Jaffe procedures was similar for acetone
and pyruvate but differed markedly for acetoacetate
and ascorbate.
For enzymatic procedures, the magnitude of the
interference observed in the supplementation studies
was generally smaller than for the Jaffe procedures, but
the pattern of interference across procedures was more
complicated. For acetoacetate, the interference was
positive for 1 enzymatic procedure, negative for 1, and
absent for the other 2. No interference was observed for

Specificity of Creatinine Measurement Procedures

acetone with any of the enzymatic procedures. For

ascorbate, 2 enzymatic procedures showed negative interference at low creatinine concentrations and 2
showed positive interference at high creatinine concentrations. Procedure E3 showed negative interference at
low creatinine concentrations and positive interference
at high creatinine concentrations. For pyruvate, 2 procedures showed positive interferences at high creatinine concentrations.
Cobbaert et al. (9 ) examined the effects of albumin, hemoglobin A, IgG, and unconjugated bilirubin
on enzymatic and Jaffe creatinine procedures by supplementing these substances into serum pools. Those
authors found positive creatinine biases from albumin
and IgG for Jaffe procedures but not for enzymatic procedures. An important difference in our experimental
design was that we selected individual clinical samples
containing the potentially interfering substances.
However, biases may have been due to presence of the
substance used to select the samples or to unknown
substances associated with samples in that disease category. We found that 2 of 3 Jaffe procedures exhibited a
positive bias with high protein samples and a few samples had a negative bias with 1 Jaffe procedure; we also
observed a positive bias with 1 enzymatic procedure.
Our findings for patient samples selected for low albumin concentrations showed that 1 Jaffe procedure had
a positive bias with 2 samples, 1 Jaffe procedure had a
negative bias with 2 samples, and 1 Jaffe procedure had
a negative bias with 1 sample and a positive bias with 7
samples. For the high total protein sample category, 1
enzymatic and 2 Jaffe procedures showed positively biased creatinine values. Although we were not able to
specifically identify the interfering substances in the
clinical samples, our results for samples with protein
concentrations not within reference intervals demonstrated that both enzymatic and Jaffe methods were
affected and that the influence of interfering substances
was complex and dependent on the technical implementation of a measurement procedure.
Previously reported effects of bilirubin on enzymatic and Jaffe procedures for serum creatinine have
been contradictory. Cobbaert et al. (9 ) found that enzymatic and some Jaffe procedures demonstrated similar slight positive or negative interferences with unconjugated bilirubin, whereas other Jaffe procedures
exhibited 10% negative interference. Owen et al.
(10 ) measured creatinine results for 73 patient samples
with bilirubin concentrations of 5.8 56 mg/dL (99
958 mol/L) using Roche enzymatic and Jaffe procedures and an LC-IDMS procedure. They found that
49% of creatinine results had 10% to 35% bias for
the enzymatic procedure and, for the Jaffe procedure, 1
result had 10% bias and 2 results had 10% bias.
For the high bilirubin patient group in our study, all

but 1 sample had bilirubin concentration 19 mg/dL

(325 mol/L), and most of these samples were collected from only 3 patients. For 3 of 4 enzymatic procedures and 1 of 3 Jaffe procedures, a large portion of
the samples were negatively biased. Interestingly, procedure E2 had only 1 biased result and procedure J1
had no biased results with the increased bilirubin samples. As for protein, our results demonstrated that the
influence of bilirubin was complex and dependent on
the technical implementation of a measurement
procedure.
Delayed separation of serum from cellular components following specimen collection has been reported
to cause increased creatinine results with some Jaffe
creatinine procedures, likely owing to accumulation of
pyruvate (11 ). For the delayed sample separation category, all 3 Jaffe procedures showed a positive bias with
variable numbers of samples. When serum pools were
supplemented with pyruvate, all 3 Jaffe procedures
demonstrated positive biases at a creatinine concentration of 0.9 mg/dL (80 mol/L) but almost no bias at
creatinine 3.8 mg/dL (336 mol/L); 2 of 4 enzymatic
procedures had positive biases only at the higher creatinine concentration.
Highly procedure-dependent interferences were
observed for the aggregate of 3 sample categories with
59 diabetes patient samples known to have increased
-hydroxybutyrate (n 19), glucose (n 20), and
Hb A1c (n 20). All 3 Jaffe procedures had a large
number of positively biased results for the glucose category. Procedures J1 and J3 had a large number of positively biased results for the -hydroxybutyrate category, whereas procedure J2 had a smaller number of
predominantly negative biases. Procedure J3 had a
large number of positive biases for the increased Hb A1c
category, and the other Jaffe procedures had essentially
no biases. For the enzymatic procedures, there were
0 3 samples with biases in each of the diabetes sample
categories, suggesting minimal or no specificity issues
for these categories. Our data do not identify the root
cause for a given bias (e.g., Hb A1c is likely not the root
cause of the bias in the increased Hb A1c patient group),
but these disease categories likely included additional
substances that influenced some creatinine procedures. For patients with diabetes, enzymatic procedures appeared to be more suitable than Jaffe
procedures.
In the cardiovascular disease group, only the J3
procedure showed positive biases. For the kidney disease patient category, none of the procedures had biased results for the sample categories predialysis, low
estimated glomerular filtration rate (eGFR), and high
urine protein where creatinine concentrations were
higher. For the post kidney transplant group with
lower creatinine concentrations, only procedure J3 had
Clinical Chemistry 58:2 (2012) 399

a few positively biased results. For the patient groups

for which creatinine is an important biomarker, it appears that properly implemented enzymatic or Jaffe
procedures gave satisfactory results.
Lidocaine showed large positive biases in a large
number of samples for all enzymatic methods, whereas
Jaffe methods had modest biases in a few samples.
There was no influence from dopamine. Dobutamine
showed positive biases with 2 of 3 Jaffe procedures and
negative biases with all enzymatic procedures. Cephalosporin showed positive biases for 2 of 3 Jaffe procedures, with no influence on the enzymatic procedures.
REVIEW OF FINDINGS VS MANUFACTURERS LABELING

We compared the biases for samples containing known

amounts of supplemented interfering substances to interference claims obtained from each manufacturers
instructions for use (IFU). In general, the IFUs did not
contain adequate information regarding the effects of
interfering substances on the procedures. Acetoacetate,
acetone, ascorbate, and pyruvate are well-known interferents in creatinine measurements (12 ). Nonetheless,
only 4 of 7 IFUs had interference claims for acetoacetate, none had claims for acetone, 5 had claims for
ascorbate, and 2 had claims for pyruvate. Of those that
did have claims for these interferences, most did not
have adequate information to interpret the claims. In
many cases, the IFUs did not state the concentrations of
interfering substances and/or creatinine concentrations tested. In some cases in which the concentration
of interferent was stated, it was substantially lower than
that expected to be encountered in diseased patient
populations (1217 ). In addition, the criteria used for
evaluation of the effects of interferences were not uniform among manufacturers.
STRENGTHS AND LIMITATIONS

Strengths of this investigation were inclusion of a substantial number of individual patient samples representing diverse clinical conditions selected to have a
high probability to contain potentially interfering substances and a control group of samples from healthy
individuals. This approach eliminated any noncommutability artifacts from influencing results and included a range of both exogenous pharmaceutical and
endogenous metabolic substances. Several labile metabolic substances were examined by supplementing a
single donor serum to ensure the substances were present in the samples tested. The LC-IDMS DCMP and
the commercial measurement procedures were validated to have calibration traceable to JCTLM listed reference measurement procedures. IDMS technology is
considered the best available to be free from the influence of interfering substances. All measurements were
400 Clinical Chemistry 58:2 (2012)

made in quadruplicate to minimize the influence of

measurement imprecision.
Limitations of this study were that we were unable
to include all manufacturers because of limited volumes available (as residual samples from clinical laboratories). The clinical samples were selected to have a
high probability to contain various interfering substances; however, the identity and concentrations of
the substances responsible for a given interference were
either unknown or only partially known based on the
selection parameters. In addition, the number of samples included in each clinical category was relatively
small (approximately 20), and in some cases, different
samples from the same individual were included more
than once in a clinical category. The clinical samples
were not handled uniformly before aliquoting and
freezing, with variable time spent at room temperature,
refrigerated, or frozen, with possible metabolic changes
or loss of labile components (e.g., dopamine and dobutamine). In addition, samples collected at ARUP
were thawed, aliquoted, and refrozen. Given previously
published findings that creatinine in serum or plasma
is a stable measurand under common sample storage
conditions (5 ), it is unlikely that sample handling or
storage contributed to underestimation of creatinine in
this study. However, the consequence of sample handling and storage variations on substances that may
interfere with creatinine measurements is unknown,
and may have led to underestimation of bias in certain
disease categories. Finally, the serum creatinine concentration interval examined in the patient samples selected for this study was inadequate to address measurement specificity issues at creatinine concentrations
typically found in pediatric patients between the ages of
2 months and 10 years, in whom values are usually
0.6 mg/dL (50 mol/L) (18 ).
Conclusions
Overall, the influence of interfering substances was less
frequent with enzymatic procedures than with Jaffe
procedures, but no procedure was unaffected by the
interfering substances or disease categories examined.
There were differences in both magnitude and direction of bias among measurement procedures within a
given method principle, enzymatic or Jaffe, indicating
that influence of interfering substances depended on
details of implementation of the method principle.
With the exception of the diabetic disease category,
which showed substantially more frequent influence of
interfering substances with Jaffe procedures than with
enzymatic procedures, no general conclusions regarding Jaffe or enzymatic technologies can be drawn. Supplemented interferents had greater influence at creatinine concentrations within reference intervals than at

Specificity of Creatinine Measurement Procedures

higher concentrations, highlighting the importance of

evaluating interference at more than 1 concentration of
the measurand. The results emphasize the need to evaluate interference characteristics in detail with each particular measurement procedure with consideration of
the patient populations served.
Manufacturers labeling and claims for interferences with commercial creatinine measurement procedures had shortcomings in the information provided.
It is recommended that manufacturers claims be based
on testing at clinically relevant concentrations of a
broad range of potential interferents as well as at 2 or
more concentrations of the measurand.
On the basis of the magnitude of biases observed in
the healthy controls category, it is recommended that
specifications for bias from interfering substances in
creatinine measurement procedures should not exceed
the larger of 0.1 mg/dL or 10% at a given concentration
of creatinine.

Author Contributions: All authors confirmed they have contributed to

the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design,
acquisition of data, or analysis and interpretation of data; (b) drafting
or revising the article for intellectual content; and (c) final approval of
the published article.

Authors Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential
Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: N. Greenberg, Ortho Clinical Diagnostics, Inc., a Johnson & Johnson Company; R.N. Dalton, SpOtOn
Clinical Diagnostics Ltd.; J.J. Zakowski, CLSI; W.G. Miller, Clinical
Chemistry, AACC, and CLSI.
Consultant or Advisory Role: None declared.
Stock Ownership: N. Greenberg, Johnson & Johnson; R.N. Dalton,
SpOtOn Clinical Diagnostics Ltd.; J.J. Zakowski, Beckman Coulter, Inc.
Honoraria: R.N. Dalton, AstraZeneca.
Research Funding: W.L. Roberts, ARUP Labs (not related to this
report); W.G. Miller (Principal Investigator), subcontract from the
University of Pennsylvania (2-U01-DK-060990-08) funded by the
NIDDK, NIH, and coordinated by the National Kidney Disease Education Program.
Expert Testimony: None declared.
Other Remuneration: R.N. Dalton, Quintiles Ltd.
Role of Sponsor: The funding organizations played no role in the
design of study, choice of enrolled patients, review and interpretation
of data, or preparation or approval of manuscript.
Acknowledgments: We thank NIST for providing SRM 967 Creatinine in Frozen Human Serum, used as a control in all assay runs. We
thank the participating manufacturers for their support by performing the measurements in their laboratories, and we especially thank
A. Hawayek and S. Wolf at Beckman Coulter; R. Brookes, S. NealonMerulla, and A. Tweedie at Ortho Clinical Diagnostics; M. Swartzentruber, B. Wiler, and F. Stubbe at Roche Diagnostics; and S.T. Salyer
at Siemens Healthcare Diagnostics for their efforts in managing the
analytical studies performed at their respective company laboratories. We also thank P. Vo at VCU and J. Hunsaker at ARUP Laboratories for their assistance with collecting residual patient samples.

References
1. Myers GL, Miller WG, Coresh J, Fleming J, Greenberg N, Greene T, et al. Recommendations for improving serum creatinine measurement: a report of
the National Kidney Disease Education Program Lab
Working Group. Clin Chem 2006;52:518.
2. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N,
Roth D, et al. A more accurate method to estimate glomerular filtration rate from serum
creatinine: a new prediction equation. Ann Intern
Med 1999;130:46170.
3. Levey AS, Coresh J, Greene T, Marsh J, Stevens
LA, Kusek JW, Van Lente F. Expressing the Modification of Diet in Renal Disease study equation
for estimating glomerular filtration rate with
standardized serum creatinine values. Clin Chem
2007;53:766 72.
4. College of American Pathologists creatinine sccuracy calibration verification/linearity survey, 2010
LN24-A participant summary. Northfield (IL): College of American Pathologists.
5. Donnelly JG, Soldin SJ, Nealon DA, Hicks JM.
Stability of twenty-five analytes in human serum
at 22 degrees C, 4 degrees C, and 20 degrees
C. Pediatr Pathol Lab Med 1995;15:869 74.

6. Preiss DJ, Godber IM, Lamb EJ, Dalton RN, Gunn

IR. The influence of a cooked-meat meal on estimated glomerular filtration rate. Ann Clin
Biochem 2007;44:35 42.
7. Dodder NG, Tai SS-C, Sniegoski LT, Zhang NF,
Welch MJ. Certification of creatinine in a human
serum reference material by GC-MS and LC-MS.
Clin Chem 2007;53:1694 9.
8. Joint Committee for Traceability in Laboratory
Medicine (JCTLM) Database of Higher-order Reference Materials, Measurements methods/procedures and services. http://www.bipm.org/jctlm/
(Accessed 13 Dec 2010).
9. Cobbaert CM, Baadenjuijsen H, Weykamp CW.
Prime time for enzymatic creatinine methods in
pediatrics. Clin Chem 2009;55:549 58.
10. Owen LJ, Keevil BG. Does bilirubin cause interference in Roche creatinine methods? Clin Chem
2007;53:370 1.
11. Shepherd J, Warner MH, Kilpatrick ES. Stability of
creatinine with delayed separation of whole
blood and implications for eGFR. Ann Clin
Biochem 2007;44:384 7.
12. Gerard SK, Khayam-Bashi H. Characterization of

13.

14.

15.

16.

17.

18.

creatinine error in ketotic patients. AJCP 1985;

84:659 64.
Weber JA, Van Zanten AP. Interferences in current methods for measurements of creatinine.
Clin Chem 1991;37:695700.
Bowers LD. Kinetic serum creatinine assays I. The
role of various factors in determining specificity.
Clin Chem 1980;26:551 4.
Bowers LD, Wong ET. Kinetic serum creatinine
assays II. A critical evaluation and review. Clin
Chem 1980;26:555 61.
CLSI. Interference testing in clinical chemistry:
approved guidelinesecond edition. CLSI document EP7-A2. Wayne (PA): CLSI; 2005.
Sacks DB. Carbohydrates. In: Burtis CA, Ashwood
ER, eds. Tietz textbook of clinical chemistry. 3rd
ed. Philadelphia: WB Saunders; 1999. p 790.
Ceriotti F, Boyd JC, Klein G, Henny J, Queralto J,
Kairisto V, Panteghini M, on behalf of the IFCC
Committee on Reference Intervals and Decision
Limits (C-RIDL). Reference intervals for serum creatinine concentrations: assessment of available
data for global application. Clin Chem 2008;54:
559 66.

Clinical Chemistry 58:2 (2012) 401